Hello everyone I was wondering what is the proper way to setup the hardware controller.
We usually setup Proxmox on Dell servers with the PERC H755 Hardware Raid controller which can be configured with different read/write policies,
Usually we configure our storage on raid 5 or raid 10 virtual disks, and we set the controller's read policy to Read Ahead and the write Policy to Write Back,
The PVE runs multiple kinds of VMs, dbs, file servers, web servers, and terminal servers, the vm disks are LVM thinpools, and the guests run ext4 or ntfs depending if they're linux or windows.
As I understand it, the Read Ahead policy is better for sequential reads, but how does that translate on the disk when there's 10 or more VMs running different kinds of applications running on the Host PVE?
As I see it there's no way to tell if the reads on the disk are sequential or random, would setting the policy to no read ahead improve performance by reducing the workload on the controller?
Also I was reading the manual for the DELL PERC 11 Series and there it said that:
"Virtual disks consisting of SSDs may not see a difference in performance using controller cache and may benefit by Fastpath."
and then:
"To enable FastPath on a virtual disk, the cache policies of the RAID controller must be set to write-through and no read ahead.
This enables FastPath to use the proper data path through the controller based on command (read/write), I/O size, and RAID
type."
I never heard of this "FastPath" technology, I tried looking it up but I didn't find much, does Proxmox Support / Implement it somehow?
Also, do these changes even matter with my current setup?
I'm not running zfs, or ceph or anything else, this is just a filesystem on lvm on a raid of SSDs,
would changing any of these policies greatly impact the performance of the VMs?
We usually setup Proxmox on Dell servers with the PERC H755 Hardware Raid controller which can be configured with different read/write policies,
Usually we configure our storage on raid 5 or raid 10 virtual disks, and we set the controller's read policy to Read Ahead and the write Policy to Write Back,
The PVE runs multiple kinds of VMs, dbs, file servers, web servers, and terminal servers, the vm disks are LVM thinpools, and the guests run ext4 or ntfs depending if they're linux or windows.
As I understand it, the Read Ahead policy is better for sequential reads, but how does that translate on the disk when there's 10 or more VMs running different kinds of applications running on the Host PVE?
As I see it there's no way to tell if the reads on the disk are sequential or random, would setting the policy to no read ahead improve performance by reducing the workload on the controller?
Also I was reading the manual for the DELL PERC 11 Series and there it said that:
"Virtual disks consisting of SSDs may not see a difference in performance using controller cache and may benefit by Fastpath."
and then:
"To enable FastPath on a virtual disk, the cache policies of the RAID controller must be set to write-through and no read ahead.
This enables FastPath to use the proper data path through the controller based on command (read/write), I/O size, and RAID
type."
I never heard of this "FastPath" technology, I tried looking it up but I didn't find much, does Proxmox Support / Implement it somehow?
Also, do these changes even matter with my current setup?
I'm not running zfs, or ceph or anything else, this is just a filesystem on lvm on a raid of SSDs,
would changing any of these policies greatly impact the performance of the VMs?